Table 3.

Challenges and possible solutions to the realisation of TCP CC with respect to 5G main mmWave network functions.

mmWave 5G Network Function	Reasons for Implementation in 5G Networks	Realisation Challenges in the 5G Network	Realisation Challenge Concerning TCP Performance	A Possible Solution to the Challenge
Frequent horizontal and vertical handovers	Reduces outage occurrence and blockage occurrence. Improves network energy efficiency, UE signal strength or BS capacity.	A lot of temporary disconnections and connections may happen in the network.	Frequent handovers can confuse the TCP when scaling its congestion window size. This reduces the capability of TCP to ensure low packet drops.	Ensuring appropriate throughput levels and TCP CC through the development of optimal handover algorithms. Implementation of devices with multi NICs and CC with multipath TCP protocols.
Usage of high frequencies in the mmWave spectrum	Transmission at higher frequencies ensures higher throughputs.	A blockage occurs since high frequencies (in mmWave spectrum) cannot pass through obstacles.	Blockages can cause the frequent triggering of TCP RTOs, longer RTTs and increase the probability of packet losses. When compared with moving UEs, these negative effects can be more evident for static UEs, since moving UEs have a faster chance of reconnection with gNB or UE.	Extending the LOS areas of the network. Putting in wireless relays in order to keep the LOS communication and optimal allocation and densification of heterogeneous network elements composed of BSs differing in size and capacity. Using intelligent reflective surfaces.
Usage of beamforming for transmission of signals from BS to UEs	Improves the coverage and signal quality by focusing the powerful signals toward a particular device.	Mismatch between the beams of the transmitter and receiver that reduce or completely eliminate the possibility of connection. Mismatched beams can cause long or short interruptions that can impact the performance of TCP.	Prevent TCP from establishing reliable end-to-end connections. High end-to-end throughput degradation in the case of NLOS communication since SNR at the location of UE cannot reach the expected values. The longer interruptions have a stronger impact on TCP performance due to the higher probability of triggering the RTOs, which further initialises the congestion window and slows the sending rate.	Development of advanced beamforming algorithms and beam tracking concepts. Implementing beam sweeping techniques, which tend to establish communication pairs after beam mismatch occurs.
Implementation of a 5G core network to support: service-based architecture, network slicing, SDN/NFV concepts	Ensures data transmission between different parts of radio access networks through a core 5G network. Ensures the realisation of a stand-alone 5G network.	Ensuring the parallel and isolated functionality of different services. Enabling appropriate separation among the different network slices. Implementation of user and data planes in separate SDN/NFVs.	TCP end-to-end congestion and flow control issues due to: the large number of simultaneously supported services, the existence of a huge number of different network slices, the separation of data and user plane traffic.	For services with high data rates, high-speed TCP CC algorithms can be used. For delay-sensitive services, the appropriate TCP CC algorithms can be deployed. Separation of the control- and user-plane with the optimal selection of distinct TCPs for each one. Implementation of QUIC protocol for CC of multiplexed web streams in core networks.
Implementation of buffering for radio link control	Enables the compensation of packet losses for higher-layer protocols.	An optimal algorithm for the selection of the buffer size. The optimal selection of the buffer location.	Implementing large buffers can cause long TCP queues. Long waiting by the packets in buffers leads to bufferbloat problems and higher latencies. Implementing small buffers decreases latency but in the case of high channel variations, an increase in dropped packets can occur. Such an increase in the number of packet losses due to reduced buffer size strongly affects loss-based TCPs.	Development of new techniques that will ensure a trade-off between performance and latency. Implementation of adjusted AQM techniques such as CoDel [57] and Flow Queue CoDel [84].
Constantly transmitted signals	Signals that enable base station detection, system information broadcasts, channel estimation, etc.	Constantly transmitted signals are independent of the UEs traffic. Such transmission consumes a part of the network capacity and negatively impacts on the energy consumption of the network devices (BSs)	Constantly transmitted signals contribute to the increase of redundant traffic and network interference. This affects the TCP performance in terms of CC and the fair distribution of data flows among the users.	Implementation of 5G networks based on the ultra-lean design based on smart signalling exchange. The ultra-lean design can reduce traffic and congestion events, and improve the TCP functionality.
Using edge computing with the support of network slicing	Reduction of the network latency through the optimised allocation of computing resources.	Large delays in the 5G network negatively impact the TCP’s functionality, especially those based on loss-based TCP protocols.	Implementing separate TCP algorithms in each slice to ensure optimal CC. Part of the applications deployed in the user-plane between the core and access network.	Solutions concerning the allocation of the servers close to UEs using approaches based on content delivery network (CDN). Development of novel TCP CC algorithms customised to the needs of a specific network slice.